BMP signaling turns up in fragile X syndrome: FMRP represses BMPR2

See allHide authors and affiliations

Science Signaling  07 Jun 2016:
Vol. 9, Issue 431, pp. fs12
DOI: 10.1126/scisignal.aaf9571


Fragile X syndrome is the most common inherited form of intellectual disability and results from a loss of function of the translational repressor FMRP. In this issue of Science Signaling, Kashima et al. find that FMRP binds to and represses a specific isoform of BMPR2, a type II bone morphogenetic protein (BMP) receptor. Reducing signaling through this BMP pathway reverses neuroanatomical defects observed in fragile X models.

Fragile X syndrome (FXS) is the most common inherited form of intellectual disability and is caused by a loss of function of the fragile X mental retardation protein (FMRP). Individuals with FXS display similarities to individuals with autism spectrum disorder, including reduced eye contact, increased repetitive behaviors, and reduced social interactions, leading to inclusion of FXS within the autism spectrum. Loss of FMRP function results in widespread alterations in neuronal morphology, synaptic function, and plasticity (1). In mouse models and in humans, loss of FMRP leads to increased numbers of immature dendritic spines (2), suggesting that defects in synapse dynamics or maturation underlie the neurological impairments (Fig. 1).

Fig. 1 Loss of FMR1 leads to increased BMPR2 abundance.

(A) In wild-type neurons, FMRP represses BMPR2, which limits LIMK1 activation and promotes actin depolymerization. (B) In FMRP mutant neurons, BMPR2 levels increase, which drives LIMK1 activation and reduces actin depolymerization.


FMRP, encoded by the FMR1 gene, is an RNA binding protein that functions as a translational repressor, leading to the straightforward hypothesis that dysregulation of its mRNA targets is responsible for neuronal phenotypes in FXS. Accordingly, intensive research has focused on the identification of mRNA binding partners of FMRP. A number of these targets encode proteins in neurotransmitter pathways, underscoring the close relationship between FMRP activity and synaptic function. However, the extensive defects in neuronal morphology observed in FXS raise the possibility that additional FMRP targets remain to be uncovered.

The current study from the Hata laboratory was motivated by their longstanding interest in bone morphogenetic protein (BMP) signaling. BMP signal transduction requires type I and II receptors, which, in turn, promote phosphorylation and activation of SMAD transcription factors. A particular type II BMP receptor, BMPR2, is unique among all BMP receptors because it contains a long C-terminal domain (CTD) dispensable for canonical SMAD-mediated signaling. Instead, this domain binds Lim kinase 1 (LIMK1) and mediates actin remodeling via LIMK1 regulation of cofilin (3, 4). In mammals, alternative splicing of the BMPR2 locus gives rise to both a 1038–amino acid full-length (FL) isoform and a 530–amino acid isoform lacking the LIMK1-interacting CTD (ΔCTD).

Kashima et al. (5) noticed that despite similar mRNA expression, the abundance of ΔCTD protein is much greater than that of FL BMPR2 in various tissues, suggesting regulation at the posttranscriptional level. Through exacting detective work, they found that the sequences encoding the CTD are responsible for inhibiting translation of the transcript encoding FL BMPR2. The authors considered possible molecular mechanisms to explain the translational repression of FL BMPR2. A key clue came in the identification of eight putative FMRP binding motifs within the FL BMPR2 transcript, including five within the CTD, suggesting that the FL BMPR2 transcript is repressed by FMRP (Fig. 1A). BMPR2 had previously emerged on lists of potential FMRP target genes (6, 7), lending further support to the hypothesis that BMPR2 could be an FMRP target. Kashima et al. went on to demonstrate that FMRP bound to the predicted binding motifs within the CTD, and that deletion of the motifs nearly eliminated the interaction between FMRP and BMPR2 mRNA and led to increased abundance of FL BMPR2. The abundance of FL BMPR2 was also markedly increased in brains of Fmr1 mutant mice, arguing that BMPR2 is an in vivo target of FMRP.

What is the relevance of FMRP-mediated repression of BMPR2 in neurons? To investigate the functional relationship between these two proteins, the authors tested whether they genetically interact in Drosophila and mouse models. If FMRP normally represses BMPR2 activity, genetically reducing BMPR2 dosage in an FMRP mutant background is predicted to reverse FMRP-dependent phenotypes. The BMPR2 homolog in flies is known as wishful thinking (Wit). Like its mammalian counterpart, Wit contains a long CTD that interacts with LIMK1 to regulate actin dynamics (8). The authors found dominant genetic interactions between dFMR1 and wit, consistent with the hypothesis that dFMRP represses Wit activity to regulate growth and morphology of the neuromuscular junction. Kashima et al. then tested whether Fmr1 and BMPR2 genetically interact in mammals. In the dentate gyrus, loss of FMRP results in a higher frequency of long, immature dendritic spines and increased overall spine density (9). Kashima et al. found that loss of one copy of BMPR2 completely rescued the spine maturation defect observed in Fmr1 homozygous knockout (KO) mice. Heterozygosity for BMPR2 did not suppress the increase in overall spine density observed in Fmr1 KO mice, suggesting that spine maturation and density are differentially sensitive to FMRP-mediated regulation of BMPR2.

If loss of FMRP results in increased abundance of FL BMPR2 and subsequent overactivation of LIMK1 (Fig. 1B), then reducing LIMK1 activity is also expected to reverse Fmr1-dependent phenotypes. The authors found that injecting LIMKi-3, a small-molecule inhibitor of LIMK1, into brains of Fmr1 KO mice potently rescued both the spine density and maturation defects in granule neurons in the dentate gyrus. Thus, pharmacological and genetic evidence indicates that inhibiting signaling through BMPR2-LIMK1 reverses dendrite morphology phenotypes associated with FXS. Last, the authors investigated whether the BMPR2-LIMK1 axis is dysregulated in human FXS patients. Remarkably, the abundance of both FL BMPR2 and phosphorylated cofilin was increased in prefrontal cortex tissue from FXS patients, arguing that FMRP also represses BMPR2 translation in human neurons.

This study describes an exciting new link between FXS and BMP signaling and draws attention to emerging functions of BMP signaling in the brain. BMPs signal at synapses and are evolutionarily conserved regulators of neuronal morphology and function (10, 11). Although BMPs can engage SMAD-dependent transcriptional pathways, they also use noncanonical pathways in neurons, including the BMPR2-LIMK1-cofilin pathway analyzed by Kashima et al. Their findings implicate BMPR2-mediated regulation of LIMK1 in neuroanatomical defects observed in FXS (Fig. 1B). To elucidate the nature of these defects, it is key to better define the role of BMP signaling in synapse formation and maturation in mammalian neurons. Moreover, it will be essential to investigate whether reducing BMPR2-LIMK1 signaling also rescues defects in synaptic function and behavior observed in FXS models. Regardless of the precise relationship between aberrant neuronal morphology and function in FXS, the discovery that FMRP controls activity of a noncanonical BMP pathway is an important step forward in deciphering molecular pathways dysregulated in FXS.


Funding: The Broihier laboratory is funded by RO1NS095895, R21NS084202, and R21NS090369.

Stay Connected to Science Signaling

Navigate This Article